[Do chronic diseases prevent intra-hospital mortality? Paradoxes and biases in information about hospital morbidity]

External Validation and Recalibration of a Mortality Prediction Model for Patients with Ischaemic Stroke

BACKGROUND

Stroke is a highly prevalent disease that can provoke severe disability. We evaluate a predictive model based on the Minimum Basic Data Set (MBDS) compiled by the Spain Health Ministry, obtained for the period 2008-2012 for patients with ischaemic stroke in Spain, to establish the model's validity and to optimise its calibration. The MBDS is the main clinical-administrative database for hospitalisations recorded in Spain, and to our knowledge, no predictive models for stroke mortality have previously been developed using this resource. The main study aim is to perform an external validation and recalibration of the coefficients of this predictive model with respect to a chronologically later cohort.

MATERIAL AND METHODS

External validation (testing the model on a different cohort to assess its performance) and recalibration (validation with optimisation of model coefficients) were performed using the MBDS for patients admitted for ischaemic stroke in the period 2016-2018. A cohort study was designed, in which a recalibrated model was obtained by applying the variables of the original model without their coefficients. The variables from the original model were then applied to the subsequent cohort, together with the coefficients from the initial model. The areas under the curve (AUC) of the recalibration and the external validation procedure were compared.

RESULTS

The recalibrated model produced an AUC of 0.743 and was composed of the following variables: age (odds ratio, OR:1.073), female sex (OR:1.143), ischaemic heart disease (OR:1.192), hypertension (OR:0.719), atrial fibrillation (OR:1.414), hyperlipidaemia (OR:0.652), heart failure (OR:2.133) and posterior circulation stroke (OR: 0.755). External validation produced an AUC of 0.726.

CONCLUSIONS

The recalibrated clinical model thus obtained presented moderate-high discriminant ability and was generalisable to predict death for patients with ischaemic stroke. Rigorous external validation slightly decreased the AUC but confirmed the validity of the baseline model for the chronologically later cohort.

Structural Model of Biomedical and Contextual Factors Predicting In-Hospital Mortality due to Heart Failure

Background: Among the clinical predictors of a heart failure (HF) prognosis, different personal factors have been established in previous research, mainly age, gender, anemia, renal insufficiency and diabetes, as well as mediators (pulmonary embolism, hypertension, chronic obstructive pulmonary disease (COPD), arrhythmias and dyslipidemia). We do not know the role played by contextual and individual factors in the prediction of in-hospital mortality. Methods: The present study has added hospital and management factors (year, type of hospital, length of stay, number of diagnoses and procedures, and readmissions) in predicting exitus to establish a structural predictive model. The project was approved by the Ethics Committee of the province of Almeria. Results: A total of 529,606 subjects participated, through databases of the Spanish National Health System. A predictive model was constructed using correlation analysis (SPSS 24.0) and structural equation models (SEM) analysis (AMOS 20.0) that met the appropriate statistical values (chi-square, usually fit indices and the root-mean-square error approximation) which met the criteria of statistical significance. Individual factors, such as age, gender and chronic obstructive pulmonary disease, were found to positively predict mortality risk. Isolated contextual factors (hospitals with a greater number of beds, especially, and also the number of procedures performed, which negatively predicted the risk of death. Conclusions: It was, therefore, possible to introduce contextual variables to explain the behavior of mortality in patients with HF. The size or level of large hospital complexes, as well as procedural effort, are key contextual variables in estimating the risk of mortality in HF.

Comparison of Three Comorbidity Measures for Predicting In-Hospital Death through a Clinical Administrative Nacional Database

BACKGROUND

Various authors have validated scales to measure comorbidity. However, the prognosis capacity variation according to the comorbidity measurement index used needs to be determined in order to identify which is the best predictor.

AIMS

To quantify the differences between the Charlson (CCI), Elixhauser (ECI) and van Walraven (WCI) comorbidity indices as prognostic factors for in-hospital mortality and to identify the best comorbidity measure predictor.

METHODS

A retrospective observational study that included all hospitalizations of patients over 18 years of age, discharged between 2017 and 2021 in the hospital, using the Minimum Basic Data Set (MBDS). We calculated CCI, ECI, WCI according to ICD-10 coding algorithms. The correlation and concordance between the three indices were evaluated by Spearman's rho and Intraclass Correlation Coefficient (ICC), respectively. The logistic regression model for each index was built for predicting in-hospital mortality. Finally, we used the receiver operating characteristic (ROC) curve for comparing the performance of each index in predicting in-hospital mortality, and the Delong method was employed to test the statistical significance of differences.

RESULTS

We studied 79,425 admission episodes. The 54.29% were men. The median age was 72 years (interquartile range [IQR]: 56-80) and in-hospital mortality rate was 4.47%. The median of ECI was = 2 (IQR: 1-4), ICW was 4 (IQR: 0-12) and ICC was 1 (IQR: 0-3). The correlation was moderate: ECI vs. WCI rho = 0.645, p < 0.001; ECI vs. CCI rho = 0.721, p < 0.001; and CCI vs. WCI rho = 0.704, p < 0.001; and the concordance was fair to good: ECI vs. WCI Intraclass Correlation Coefficient type A (ICC_A) = 0.675 (CI 95% 0.665-0.684) p < 0.001; ECI vs. CCI ICC_A = 0.797 (CI 95% 0.780-0.812), p < 0.001; and CCI vs. WCI ICC_A = 0.731 (CI 95% 0.667-0.779), p < 0.001. The multivariate regression analysis demonstrated that comorbidity increased the risk of in-hospital mortality, with differences depending on the comorbidity measurement scale: odds ratio [OR] = 2.10 (95% confidence interval [95% CI] 2.00-2.20) p > |z| < 0 using ECI; OR = 2.31 (CI 95% 2.21-2.41) p > |z| < 0 for WCI; and OR = 2.53 (CI 95% 2.40-2.67) p > |z| < 0 employing CCI. The area under the curve [AUC] = 0.714 (CI 95% 0.706-0.721) using as a predictor of in-hospital mortality CCI, AUC = 0.729 (CI 95% 0.721-0.737) for ECI and AUC = 0.750 (CI 95% 0.743-0.758) using WCI, with statistical significance (p < 0.001).

CONCLUSION

Comorbidity plays an important role as a predictor of in-hospital mortality, with differences depending on the measurement scale used, the van Walraven comorbidity index being the best predictor of in-hospital mortality.

Predictive Model and Mortality Risk Score during Admission for Ischaemic Stroke with Conservative Treatment

Background: Stroke is the second cause of mortality worldwide and the first in women. The aim of this study is to develop a predictive model to estimate the risk of mortality in the admission of patients who have not received reperfusion treatment. Methods: A retrospective cohort study was conducted of a clinical–administrative database, reflecting all cases of non-reperfused ischaemic stroke admitted to Spanish hospitals during the period 2008–2012. A predictive model based on logistic regression was developed on a training cohort and later validated by the “hold-out” method. Complementary machine learning techniques were also explored. Results: The resulting model had the following nine variables, all readily obtainable during initial care. Age (OR 1.069), female sex (OR 1.202), readmission (OR 2.008), hypertension (OR 0.726), diabetes (OR 1.105), atrial fibrillation (OR 1.537), dyslipidaemia (0.638), heart failure (OR 1.518) and neurological symptoms suggestive of posterior fossa involvement (OR 2.639). The predictability was moderate (AUC 0.742, 95% CI: 0.737–0.747), with good visual calibration; Pearson’s chi-square test revealed non-significant calibration. An easily consulted risk score was prepared. Conclusions: It is possible to create a predictive model of mortality for patients with ischaemic stroke from which important advances can be made towards optimising the quality and efficiency of care. The model results are available within a few minutes of admission and would provide a valuable complementary resource for the neurologist.

Predictive Model of the Risk of In-Hospital Mortality in Colorectal Cancer Surgery, Based on the Minimum Basic Data Set

BACKGROUND

Various models have been proposed to predict mortality rates for hospital patients undergoing colorectal cancer surgery. However, none have been developed in Spain using clinical administrative databases and none are based exclusively on the variables available upon admission. Our study aim is to detect factors associated with in-hospital mortality in patients undergoing surgery for colorectal cancer and, on this basis, to generate a predictive mortality score.

METHODS

A population cohort for analysis was obtained as all hospital admissions for colorectal cancer during the period 2008-2014, according to the Spanish Minimum Basic Data Set. The main measure was actual and expected mortality after the application of the considered mathematical model. A logistic regression model and a mortality score were created, and internal validation was performed.

RESULTS

115,841 hospitalization episodes were studied. Of these, 80% were included in the training set. The variables associated with in-hospital mortality were age (OR: 1.06, 95%CI: 1.05-1.06), urgent admission (OR: 4.68, 95% CI: 4.36-5.02), pulmonary disease (OR: 1.43, 95%CI: 1.28-1.60), stroke (OR: 1.87, 95%CI: 1.53-2.29) and renal insufficiency (OR: 7.26, 95%CI: 6.65-7.94). The level of discrimination (area under the curve) was 0.83.

CONCLUSIONS

This mortality model is the first to be based on administrative clinical databases and hospitalization episodes. The model achieves a moderate-high level of discrimination.

Suicidal behaviour characteristics and factors associated with mortality in the hospital setting

INTRODUCTION

Suicide is a major public health problem worldwide, and an approach is necessary due to its high potential for prevention. This paper examines the main characteristics of people admitted to hospitals in the Community of Madrid (Spain) with suicidal behaviour, and the factors associated with their hospital mortality.

MATERIAL AND METHODS

A study was conducted on patients with E950-E959 codes of suicide and self-inflicted injuries of the International Classification of Diseases, Ninth Revision, Clinical Modification, contained in any diagnostic field of the minimum basic data set at hospital discharge between 2003 and 2013. Sociodemographic, clinical and health care variables were assessed by uni- and multivariate logistic regression analysis in the evaluation of factors associated with hospital mortality.

RESULTS

Hospital suicidal behaviour predominates in women (58.7%) and in middle-age. Hospital mortality is 2.2% (1.6% in women and 3.2% in men), increasing with age. Mental disorders are detected 3-4 times more in secondary diagnoses. The main primary diagnosis (>74%) is poisoning with substances, with lower mortality (∼1%) than injury by hanging and jumping from high places (≥12%), which have the highest numbers. Other factors associated with increased mortality include different medical comorbidities and severity of the injury, while length of stay and mental disorders are protective factors. Type of hospital, poisoning, and Charlson index are associated differently with mortality in men and women.

CONCLUSIONS

Hospitalised suicidal acts show a low mortality, mainly related to comorbidities and the severity of injuries.

Hospital mortality in postoperative critically ill patients older than 80 years. Can we predict it at an early stage?

OBJECTIVES

To determine the incidence of in-hospital mortality throughout the post-surgical period of patients aged 80 or over who were admitted to the post-surgical critical care unit, as well as to assess the predictive capacity of those variables existing in the first 48hours on the in-hospital mortality.

MATERIAL AND METHODS

An observational retrospective cohort study conducted on postsurgical patients up to 80years old who were admitted to the unit between June 2011 and December 2013. Univariate and multivariate binary logistic regression was used to determine the association between mortality and the independent variables.

RESULTS

Of the 186 patients included, 9 (4.8%) died in the critical care unit, and 22 (11.8%) died in wards during hospital admission, giving a hospital mortality of 31 (16.7%). Among the 78 patients (42%) that underwent acute surgery, and the 108 who underwent elective surgery, there was a mortality rate of 19 (10.2%) and 12 (6.5%), respectively. As regards the variables analysed during the first 48hours of admission that showed to be hospital mortality risk factor were the need for mechanical ventilation over 48h, with an OR: 7.146 (95%CI: 1.563-32.664, P=.011) and the degree of the severity score on the APACHE II scale in the first 24hours, with an OR: 1.102 (95%CI: 1.005-1.208, P=.039).

CONCLUSION

The incidence of hospital mortality in very old patients found in our study is comparable to that reported by other authors. Patients who need mechanical ventilation over 48h, and with higher scores in the APACHE II scale could be at a higher risk of in-hospital mortality.